Regulation of the opposing (p)ppGpp synthetase and hydrolase activities in a bifunctional RelA/SpoT homologue from Staphylococcus aureus

The stringent response is characterized by (p)ppGpp synthesis resulting in repression of translation and reprogramming of the transcriptome. In Staphylococcus aureus, (p)ppGpp is synthesized by the long RSH (RelA/SpoT homolog) enzyme, RelSau or by one of the two short synthetases (RelP, RelQ). RSH enzymes are characterized by an N-terminal enzymatic domain bearing distinct motifs for (p)ppGpp synthetase or hydrolase activity and a C-terminal regulatory domain (CTD) containing conserved motifs (TGS, DC and ACT). The intramolecular switch between synthetase and hydrolase activity of RelSau is crucial for the adaption of S. aureus to stress (stringent) or non-stress (relaxed) conditions. We elucidated the role of the CTD in the enzymatic activities of RelSau. Growth pattern, transcriptional analyses and in vitro assays yielded the following results: i) in vivo, under relaxed conditions, as well as in vitro, the CTD inhibits synthetase activity but is not required for hydrolase activity; ii) under stringent conditions, the CTD is essential for (p)ppGpp synthesis; iii) RelSau lacking the CTD exhibits net hydrolase activity when expressed in S. aureus but net (p)ppGpp synthetase activity when expressed in E. coli; iv) the TGS and DC motifs within the CTD are required for correct stringent response, whereas the ACT motif is dispensable, v) Co-immunoprecipitation indicated that the CTD interacts with the ribosome, which is largely dependent on the TGS motif. In conclusion, RelSau primarily exists in a synthetase-OFF/hydrolase-ON state, the TGS motif within the CTD is required to activate (p)ppGpp synthesis under stringent conditions.

(p)ppGpp is synthesized by cytoplasmic enzymes that contain a conserved synthetase domain. RelA of Escherichia coli was the first such enzyme described and has been shown to synthesize (p)ppGpp under conditions of amino acid limitation [12]. E. coli and many other gram-negative bacteria possess an additional enzyme, SpoT, that possesses (p)ppGpp synthetase and hydrolase activities. The (p)ppGpp synthetase activity of SpoT is stimulated by various conditions, e.g. fatty acid deprivation [13,14]. In Firmicutes, homologous enzymes (Rel) constitute a distinct class of (p)ppGpp synthetases [11,15,16]. RelA, SpoT and Rel enzymes all belong to RSH (for RelA/SpoT homolog) superfamily [15,17]. Similar to SpoT, the Rel enzymes from Firmicutes are bifunctional proteins with (p)ppGpp synthetase and hydrolase activities; however, similar to RelA, the synthetase activity of these enzymes is stimulated upon amino acid starvation [18,19]. RSH enzymes share a multi-domain architecture with a C-terminal regulatory domain (CTD) and an N-terminal enzymatic domain (NTD) containing synthetase and hydrolase motifs. The only available crystal structure of an RSH enzyme is that of the NTD of Rel from Streptococcus equisimilis [20]. The structure indicates two conformations of the enzyme, corresponding to the reciprocal active states of the enzyme: (p)ppGpp-synthetase-ON/hydrolase-OFF (stringent) and synthetase-OFF/hydrolase-ON (relaxed). It has been proposed that the CTD is involved in reciprocal regulation of the enzymatic states. The current model suggests that under non-stringent (relaxed) conditions, the interaction of the CTD with the NTD maintains the enzyme in the synthetase-OFF/hydrolase-ON conformation [21,22,23] The CTD of RelA stimulates (p)ppGpp synthesis in a ribosome-dependent manner when uncharged tRNA, as a consequence of amino acid limitation, is located in the ribosomal A-site [24,25]. Interestingly, Rel from S. equisimilis is responsive to amino acid starvation only within its native genetic background and not when expressed in E. coli [13,21]. Bioinformatic analyses have revealed the presence of three conserved motifs within the CTDs of RSHs: TGS, ACT and DC. The TGS motif (named after the presence in ThrRS, GTPases, and SpoT) was shown to be responsible for the interaction of SpoT with the acyl-carrier protein. The ACT motif (named after three of the allosterically regulated enzymes in which this domain is found: aspartate kinase, chorismate mutase and TyrA) was proposed to be a conserved regulatory ligand-binding fold [26,27]. Recently, major insights into the ribosome-RelA structure were provided by cryo-EM analyses [28,29,30]. The structures revealed that RelA adopts an open conformation in which the CTD is intertwined around an A-site tRNA within the intersubunit cavity of the ribosome, and the NTD extends into the solvent. The structures support a model in which association of monomeric RelA with the ribosome relieves the autoinhibitory effect of the CTD on the NTD. It was hypothesized that autoinhibition in the unbound state is mediated by oligomerization of RelA. Oligomerization was previously demonstrated to occur via a conserved aspartate-cysteine motif (DC) in the CTD [31,32,33]. Interaction of monomeric RelA with the ribosome and putative RelA oligomerization in the unbound state indicate that the switching of enzymatic activities occurs via a complex mechanism that has not yet been elucidated.
In the human pathogen Staphylococcus aureus, the stringent response plays an important role in virulence [18], phagosomal escape [34] and antibiotic tolerance [35]. In S. aureus, in addition to Rel Sau , two enzymes with (p)ppGpp synthetase activity (RelP and RelQ) are present. These enzymes form homotetramers that lack the CTD and the hydrolase domain [36,37,38] and are transcriptionally induced under conditions of cell-wall stress [35]. The basal (p)ppGpp level produced by these enzymes is controlled by the hydrolase activity of Rel Sau [35]. The phenotypic consequences of (p)ppGpp accumulation vary among species and can be mediated by different mechanisms. In S. aureus, as in other Firmicutes, (p)ppGpp regulates transcription by an indirect mechanism that strongly relies on the lowering of intracellular GTP levels [39,40,41]. Low GTP levels lead to de-repression of the CodY regulon. CodY, when loaded with GTP and branched-chain amino acids, acts as a repressor of a variety of genes, e.g., genes involved in amino acid synthesis and virulence [41] A decrease in GTP levels could also lead to the repression of sensitive GTP-initiating promoters (e.g., those of stable RNA genes) [42,43]. All these studies illustrate the complex role of (p)ppGpp during the bacterial life cycle. The cellular concentration of (p)ppGpp has to be tightly regulated not only to support survival under stressed conditions but also to avoid toxicity under relaxed conditions. The molecular switch between the synthetase and hydrolase activities of Rel Sau is crucial for the maintenance of this balance.
Here, we aim to elucidate the role of the CTD in controlling the activity of Rel of the major human pathogen S. aureus (Rel Sau ) in vivo. We show that the (p)ppGpp synthetase activity is restricted in S. aureus and that the synthetase is activated only upon interaction of the CTD with ribosomal partners under stringent conditions. The TGS and DC motifs within the CTD are essential for the enzymatic switch to the synthetase-ON state and play a major role in the interaction between Rel Sau and the translational apparatus.

S. aureus background
We aimed to analyze the role of the CTD of Rel Sau in the stringent response in S. aureus. In Rel Sau the canonical domains and motifs could be identified through alignment with RelA and SpoT from E. coli (Fig 1A). We first established a readout system for (p)ppGpp activity. To this end, we analyzed strain HG001 (wild type) as well as an isogenic mutant of this strain that carries mutations in all three (p)ppGpp enzymes (full deletion of rel, synthetase mutation in relP and relQ) and thus is unable to synthesize pppGpp (designated (p)ppGpp 0 ) [35]. The mutant exhibits no phenotypic difference compared to the wild type when grown in rich medium ( Fig  1B). The stringent response can be evoked by mupirocin, an inhibitor of isoleucyl-tRNA synthetase [18,34]. (p)ppGpp synthesis results in higher tolerance towards mupirocin. The (p) ppGpp 0 strain exhibited a typical decline in OD 600 when treated with mupirocin ( Fig 1C). Furthermore, synthesis of (p)ppGpp results in repression of genes coding for ribosomal proteins (e.g., rpsL) and de-repression of the CodY target genes (e.g., SAOUHSC_02923, a putative amino acid transporter) ( Fig 1D). Therefore, we used the enhanced mupirocin tolerance and typical transcription pattern (rpsL down, SAOUHSC_02923 up) as a readout for (p)ppGpp synthesis in S. aureus. As a first approach, we deleted the CTD of the wild-type Rel Sau (SAOUHSC_01742) to generate strain HG001-531. In contrast to a full-length rel deletion (Geiger et al., 2014a), truncation of the CTD had only a slight effect on growth ( Fig 1C). It has been previously shown that Rel Sau is essential due to its hydrolase function [35]. Thus, the CTD does not seem to impede the hydrolase activity in vivo. Northern blot analysis revealed that the CTD mutant was unable to elicit a mupirocin-induced stringent response ( Fig 1B). The transcriptional pattern of the marker genes rpsL and SAOUHSC_02923 as well as the mupirocin tolerance (Fig 1D) of the CTD mutant were indistinguishable from those of the (p) ppGpp 0 strain. This finding is consistent with the general assumption that the CTD is required for sensing amino acid deprivation. However, hydrolase activity seems to be hardly effected by the CTD.
Next, we complemented the (p)ppGpp 0 strain with anhydrotetracycline (ATc)-inducible full-length and truncated rel constructs (Fig 2A) and analyzed the effects under relaxed growth conditions (exponential growth phase in nutrient-rich medium). As a positive control, we induced relQ expression. RelQ is a small synthetase without a regulatory CTD and thus can activate the stringent response in a (p)ppGpp 0 mutant by transcriptional induction alone. This activation was demonstrated by downregulation of rpsL and upregulation of SAOUHSC_02923 ( Fig 2B) and by the immediate growth arrest after relQ induction ( Fig 2C).
In contrast to relQ, transcriptional induction of full-length rel showed no effect on the transcription of marker genes or on growth (Fig 2B and 2D). This finding confirms that additional post-transcriptional activation is required to activate the synthetase activity. Induction of a construct lacking the CTD also failed to induce the stringent response phenotype (Fig 2B and  2D). At first glance, these results may indicate that under relaxed conditions, the enzymatic domain of Rel Sau , with or without the CTD, is tightly held in a synthetase-OFF conformation. Alternatively, the hydrolase might be hyperactive, so any (p)ppGpp synthesized would be immediately degraded. To test this hypothesis, we mutated the hydrolase domain in full-length and CTD-deleted rel constructs. Indeed, both full-length and truncated rel lacking the hydrolase domain elicited a stringent response pattern similar to that of the wild type, as indicated by transcriptional and growth analyses (Fig 2B and 2D). Thus, we presume that there might be some synthetase activity under relaxed growth conditions. However, due to hydrolase activity, any (p)ppGpp present is efficiently degraded under these conditions.

CTD of Rel Sau does not impact hydrolase activity in vivo
To analyze the hydrolase activity of Rel Sau in vivo, we used a conditional rel mutant strain (HG001-55) [18] in which genomic rel was placed under an IPTG-inducible promoter complemented with different rel constructs ( Fig 3A). Without IPTG, the rel mutant is unable to grow ( Fig 3B) because it cannot degrade the (p)ppGpp synthesized by RelP and RelQ [35]. We introduced ATc-inducible full-length or truncated rel into the conditional rel mutant and monitored growth after ATc induction. As expected, constructs with mutated hydrolase could not rescue the growth defect of HG001-55 ( Fig 3C). However, full-length and CTD-truncated rel, with intact hydrolase, fully complemented the growth defect of the conditional rel mutant. These results show that the hydrolase was constitutively active, independent of the presence of the CTD. In summary, the data indicate that under relaxed conditions, the wild-type Rel Sau enzyme, with or without sensory domain, is tightly held in the hydrolase-ON state.

Role of the CTD in the enzymatic activity of Rel Sau in vitro
To confirm the data from the in vivo experiments under relaxed conditions, full-length or CTD-truncated Rel Sau proteins (with or without hydrolase domains) were purified and tested in vitro for enzymatic activities (Fig 4A and 4B). In the synthetase reaction, pyrophosphate is transferred from ATP to GTP, yielding AMP and pppGpp. The presence of both products was measured by HPLC-MS. AMP production was detectable with all constructs ( Fig 4C); however, the AMP levels were significantly higher for the constructs that lacked the CTD (Fig 4C), indicating that the CTD negatively interferes with synthetase activity. Interestingly, pppGpp production was not detected for constructs with intact hydrolase (Fig 4D). However, proteins with mutated hydrolases synthesized detectable amounts of (p)ppGpp. The enzyme lacking the CTD showed slightly higher pppGpp synthetase activity than the full-length Rel Sau supporting the inhibitory effect of the CTD on the synthetase domain. Thus, Rel Sau exhibits strong hydrolase activity, which prevents pppGpp accumulation. This finding was confirmed by the rapid degradation of pppGpp ( Fig 4E) and ppGpp ( Fig 4F) by full-length and CTD deleted Rel Sau . Notably, Rel Sau preferentially degraded ppGpp over pppGpp. The CTD apparently has a minor impact on hydrolase activity.

Role of the CTD of Rel Sau in E. coli
Our analysis of Rel Sau in its native background seemed to be inconsistent with the results of previous studies, in which different CTD-deleted enzymes from other organisms were expressed in E. coli [20,21,44,45]. These studies indicated that RSH enzymes that lack CTDs are in a synthetase-ON/hydrolase-OFF state. Thus, based on these studies, we also expressed full-length and CTD-deleted rel in E. coli using an arabinose-inducible promoter. We tested the capacity of different rel constructs (Fig 5A) to complement the defective phenotype of MG1655, a relA/spoT mutant, under stringent conditions (Fig 5B and 5C). Full-length and CTD-deleted Rel Sau were able to complement the relA/spoT mutation. The complementation could be attributed to (p)ppGpp synthetase activity: mutation within the synthetase domain abolished complementation, whereas mutation within the hydrolase domain did not affect the complementation assay. Thus, in E. coli, CTD-deleted Rel Sau , similar to other RSH enzymes, is predominantly in a synthetase-ON/hydrolase-OFF state, whereas in S. aureus, this enzyme is primarily in a hydrolase-ON state.

Role of CTD motifs for sensing in vivo
Within the CTDs of RSHs, several conserved motifs can be identified. The conserved TGS, DC and ACT motifs of Rel Sau were predicted based on sequence alignments, and the critical residues of these motifs were mutated (Fig 1A). Wild-type and CTD-mutated rel were cloned to be under the control of the native rel promoter and introduced into the (p)ppGpp 0 strain ( Fig  6A). The stringent response upon mupirocin treatment was analyzed by Northern blotting and growth analysis (Fig 6B and 6C). A (p)ppGpp 0 strain containing the empty vector showed the typical decrease in OD 600 after mupirocin treatment. Induction of full-length rel in the (p) ppGpp 0 mutant fully complemented the mutant phenotype, whereas the CTD-deleted rel was unable to do so. Mutation of the ACT motif resulted in slightly impaired complementation. However, mutation of the TGS or DC motif resulted in complete inactivation of the stringent response. Expression of these mutated rel genes resulted in a phenotype that was not distinguishable from the phenotype of the (p)ppGpp 0 strain in terms of growth and gene expression pattern. Thus, the TGS and DC motifs are required for stringent response, while the ACT motif plays only a minor role.

Co-immunoprecipitation (Co-IP) of Rel Sau
We aimed to analyze the role of the conserved motifs within the CTD for interaction with cytosolic proteins. Therefore, we performed Co-IP experiments using whole-cell lysates of (p) ppGpp 0 mutants expressing wild-type or mutated (ACT, DC and TGS see Fig 1A) versions of Rel Sau . For each pull-down experiment, the wild-type or mutant Rel Sau was the most abundant protein detected, with no significant difference observed between wild type and mutant proteins (Data S1 Dataset) and the expression of all proteins was similar as shown by Western blot analysis (Blot in S1 Fig). Mainly ribosomal proteins were co-immuno-precipitated with native Rel Sau . When Rel Sau with mutated TGS motif was used as bait significant less proteins were enriched (Fig 7 first column). Most of these putative TGS interacting proteins were also found to be effected when Rel Sau harboring mutations in ACT or DC motifs were used, although to a lesser extent. Immuno-precipitated proteins that were strongly influenced by the TGS mutation are ribosomal proteins, proteins associated with RNA degradation and proteins involved in DNA-related pathways. In summary, the results indicate that all three motifs within the CTD work together to dock Rel Sau onto the translational apparatus. The strongest interaction is mediated by TGS, whereas the ACT motif seems to have a low impact. The TGS motif seems to mediate also interaction with non-ribosomal proteins.

Discussion
RSH enzymes are major players in the synthesis and hydrolysis of the second messenger (p) ppGpp. There is still limited information about the molecular switch that regulates the two activities, both present in long RSH enzymes. Here, we analyzed how the CTD of Rel Sau influences the enzymatic activities in vivo. We showed that Rel Sau exists primarily in a synthetase-OFF/hydrolase-ON conformation. Only under stringent growth conditions was the switching to the synthetase-ON conformation detectable, and this switching occurred only when the CTD possessed intact TGS and DC motifs.
In S. aureus and probably in other Firmicutes, Rel combines the functions of the two prototypic RSH enzymes, RelA and SpoT, from Proteobacteria. The synthetase activity is needed to elicit a stringent response phenotype, presumably via interaction with ribosomes and uncharged tRNA, as previously shown for RelA [24,25]. However, similar to SpoT, Rel Sau also possesses strong hydrolase activity, which is necessary to counteract the (p)ppGpp production by the small synthetases RelP and RelQ present in Firmicutes. The equilibrium between these two activities needs to be tightly regulated in order to attain an appropriate level of (p)ppGpp based on the growth conditions. So far, potential differences between RelA, SpoT and Rel associated with the molecular switch could not be inferred from the sequence or in vitro analyses. In vivo activities of different RSH enzymes were mainly analyzed by heterologous expression in E. coli. These analyses indicated that without CTDs, RSH enzymes possess strong synthetase activity [21,44,45]. Similarly, Rel Sau , with or without the CTD, can complement an E. coli relA/ spoT mutant, also indicating that (p)ppGpp synthesis can occur with or without the CTD. However, analysis of the same construct in the native background clearly showed that Rel Sau lacking the CTD is tightly held in the hydrolase-ON state, and synthetase activity is detectable only in constructs that lack the hydrolase. Thus, Rel Sau , with or without the CTD, exhibits net (p)ppGpp synthetase activity when expressed in E. coli but net hydrolase activity when expressed in S. aureus. It would be interesting to see whether enzymes from other organisms show a similar discrepancy between E. coli and native backgrounds. Our results indicated that the enzymatic activity of Rel Sau is influenced by species-specific interactions of the enzymatic NTD with unknown factors. To date, there is no evidence that the NTD alone interacts with the ribosome. Thus, other interaction partners or intracellular properties of the NTD should be elucidated in the future. An alternative possibility is that less (p)ppGpp is needed to complement the phenotype of a pppGpp mutant in E. coli allowing growth even if Rel Sau has a weak synthetase. However, this is not supported by our in vitro results, showing that synthetase activity is only detectable when the hydrolase is mutated in constructs with or without CTD. Analysis of full-length or truncated Rel Sau in vitro largely confirmed the results obtained with the in vivo data obtained in S. aureus. The presence of an intact hydrolase abrogates the synthetase activity. Synthesis of pppGpp was detectable only in hydrolase-deficient constructs. Moreover, we show that the CTD has an inhibitory effect on synthetase activity since truncated versions of Rel Sau showed higher accumulation of the reaction products AMP and pppGpp compared to the full-length enzyme. However, the CTD had only a minor impact on the strong hydrolase activity of the purified enzymes. Interestingly, Rel Sau preferentially hydrolyzes ppGpp over pppGpp. In vivo, it was shown that RelP and RelQ mainly produce ppGpp [35], which is toxic at high concentrations and requires efficient hydrolysis. This observation could explain the preference of Rel Sau for ppGpp hydrolysis.
The Co-IP results indicate that Rel Sau interacts with the translation machinery and that the TGS strongly influence this interaction. This is largely consistent with previous data obtained for RelA [24,25,28,29,30]. Among the 10 detected interacting ribosomal proteins, L16, S13, and S12 are homologous to E. coli proteins that have been previously identified to interact with RelA [28,29,30]. Of note, the TGS motif also seems to hamper the putative interaction of Rel Sau with other proteins of the RNA and DNA pathways. Whether such interactions are specific and involved in the molecular function of Rel Sau remains to be investigated.
The in vivo analyses combined with the Co-IP results provided some clues regarding the roles of the different motifs of the CTD of Rel Sau . Of the three motifs, the TGS motif showed the strongest effect, and the ACT motif showed the weakest effect, on (p)ppGpp activation and ribosomal interaction. Thus, the role of the ACT motif remains to be elucidated but seems to be minor. The TGS motif is clearly required for synthetase activation, most likely interacts with the ribosome to sense whether or not the tRNA in the A-site is aminoacetylated as shown for RelA [28,29,30,46]. Similar to the TGS motif, the DC motif was also found to be required for synthetase activity and to influence interactions with ribosomal proteins. DC has also been reported to interact with 23S rRNA ASF [28] which is critical to RelA activation in E. coli [46], presumably through stabilizing ribosome interaction.
This finding contradicts the simple model in which the DC motif causes oligomerization and thereby autoinhibition [31,32,33]. This would imply that DC mutation alleviates autoinhibition leading to increased synthase activity. In contrast, our data showed that the DC-mutated Rel Sau is held in a synthetase-OFF state. Thus, our data support a model in which the DC motif participates in specific activation upon ribosomal contact, and that this interaction is involved in the intramolecular switch.

Strains and growth conditions
Strains and plasmids are listed in the table in S1 Table. For strains carrying resistance genes, antibiotics (10 μg/ml erythromycin, 5 μg/ml tetracycline, 10 μg/ml chloramphenicol, and 100 μg/ml ampicillin) were used only in precultures. For the conditional mutant HG001-55, IPTG (final concentration of 0.5 mM) was added only in the preculture. S. aureus strains were grown in CYPG (10 g/l casamino acids, 10 g/l yeast extract, 5 g/l NaCl, 0.5% glucose and 0.06 M phosphoglycerate) medium [47]. Bacteria from an overnight culture were diluted to an initial optical density (OD 600 ) of 0.05 in fresh medium and grown with shaking (220 rpm) at 37˚C to the desired growth phase. Expression of cloned proteins was induced in exponential phase (OD 600 = 0.3) with 0.1 μg/ml anhydrotetracycline (ATc) and in stringent conditions by addition of 0.3 μg/ml mupirocin. E. coli strains were grown in an overnight preculture in LB medium. Stringent conditions were applied by growing cells in modified M9 medium (33.7 mM NaHPO 4 , 22 mM KH 2 PO 4 , 8.55 mM NaCl, 9.35 mM NH 4 Cl, 1 mM MgSO 4 , 0.3 mM CaCl 2 , 1 μg/ml thiamine hydrochloride, 0.4% glycerol, 1 mM serine, 1 mM methionine and 1 mM glycine) [48]. For growth on solid media, single colonies grown on LB agar were streaked on M9 agar plates. For growth curve analyses, bacteria were inoculated to the desired OD 600 (S. aureus initial OD 600 = 0.05; E. coli initial OD 600 = 0.1) in a 96-well plate, and growth was monitored in an Infinite M200 Pro microplate reader (Tecan).

Construction of vectors for expression of full-length and truncated rel
All oligonucleotides are listed in S2 Table. ATc-inducible plasmids, derived from pCG248, were generated with a restriction enzyme cloning strategy. Amplicons and vector were digested with EcoRI restriction enzyme. Substitution of the hydrolase domain and the ACT, TGS and DC mutations were achieved by overlapping PCR. For expression of the rel constructs under the native promoter, the shuttle vector pCG443 was designed based on pJL77 [49]. pJL77 was digested with AscI and SphI to remove the previous insert, including the promoter. The rel promoter was amplified from genomic DNA, digested using the same restriction enzymes, and ligated to generate pCG443. Full-length, truncated and mutated versions of rel were subcloned from the pCG248 plasmids into AscI-digested pCG443 by Gibson assembly [50].
All inserts were verified by sequencing (4base lab AG advanced molecular analysis), electroporated into the restriction-deficient S. aureus strain RN4220, and then transduced into the final S. aureus strains. All S. aureus strains were tested by PCR for the presence of the correct plasmid.
For expression in E. coli, different derivatives of rel were cloned into the EcoRI site of pBAD30 via Gibson assembly using the oligonucleotides listed in Table S2. Resulting vectors were verified by sequencing and moved to MG1655 E. coli strains (wild type and relA/spoT mutant) [51,52].
For protein purification, different rel derivatives were subcloned from the pCG248-based plasmids into BamHI-digested pET15b using Gibson assembly.

Generation of S. aureus mutant strains
The markerless rel CTD-deletion mutant was obtained using the ATc-inducible suicide vector pBASE6 [34]. Deletion was introduced by overlapping PCR with the primers listed in S2 Table, and the amplicon was cloned into BglII-and SalI-digested pBASE6 by Gibson assembly. The resulting plasmid was verified by sequencing and electroporated into RN4220, from which the plasmid was transduced into HG001. Mutagenesis was performed as described previously [34]. Mutation was verified by PCR.

RNA isolation and Northern blot analysis
RNA isolation and Northern blot analysis were performed as described previously [53]. Briefly, 5 ml of bacteria were collected at the desired time point (30 minutes after induction) and centrifuged. The pellet was resuspended in 1 ml of TRIzol reagent (Thermo Fisher Scientific) with 0.5 ml of zirconia/silica beads (0.1-mm diameter) and lysed using a high-speed homogenizer (Thermo Fisher Scientific). RNA was isolated following the instructions provided by the TRIzol manufacturer. For the detection of specific transcripts on the Northern blot, digoxigeninlabeled probes were generated using the DIG-labeling PCR Kit as described by the manufacturer (Roche Life Science).

Purification of different Rel Sau constructs
E. coli BL21 (DE3) (New England Biolabs) cells that were freshly transformed with plasmids carrying full-length rel constructs were grown for 16 hours at room temperature under constant shaking (150 rpm) in LB medium supplemented with D(+)-lactose-monohydrate (12.5 g/l) and ampicillin (100 μg/ml). Cells were harvested (20 minutes, 3000 x g, 4˚C) and resuspended in ice-cold high-KCl buffer A (20 mM HEPES (pH 7.4), 20 mM NaCl, 20 mM MgCl 2 , 1 M KCl, 30% (v/v) glycerol, and 40 mM imidazole) supplemented with 10 μg/ml DNAse and cOmplete protease inhibitor cocktail (Roche). Cells were lysed by a French press at 1000 psi. The lysate was centrifuged (50,000 x g, 45 minutes, 4˚C), and the clear supernatant was filtered (0.22-μm pore size) before being loaded onto a 1-ml HisTrap HP column (GE Healthcare Life Sciences) equilibrated with high-KCl buffer A. Purification was performed with an Ä KTA purification system (GE Healthcare Life Sciences), and elution was carried out with an imidazole gradient to a final concentration of 500 mM. Fractions were analyzed by SDS-PAGE, and the fractions containing the protein of interest were collected and concentrated to 5 ml with an Amicon Ultracel-50K ultracentrifugal device, with a cut-off of 50 kDa (Merck Millipore). Protein was further purified by size-exclusion chromatography (HiLoad 16/600 Superdex 200 pg, GE Healthcare Life Sciences). The size-exclusion column was previously equilibrated with ice-cold high-KCl SEC buffer (20 mM HEPES (pH 7.0), 20 mM NaCl, 20 mM MgCl 2 , 1 M KCl, and 30% (v/v) glycerol). Protein-containing fractions were pooled, concentrated by ultra-filtration with a 50-kDa cut-off, aliquoted and stored at -80˚C. For purification of CTD-truncated constructs, the same procedure was followed using different buffers:

In vitro assay for enzymatic activity and HPLC-MS analysis
Synthetase assays were performed in reaction buffer (20 mM HEPES (pH 7.0), 200 mM NaCl, 20 mM MgCl 2 , and 20 mM KCl) with 1 mM ATP, 1 mM GTP and 2 μM purified enzyme. Hydrolase assays were performed in the same reaction buffer with 1 mM ppGpp or 1 mM pppGpp (both from Jena Biosciences) and 0.1 μM purified enzyme. Assays were performed at 37˚C; aliquots were taken at the indicated times; and the enzyme reactions were stopped by addition of an equal volume of chloroform. The mixtures were briefly vortexed and centrifuged (3 minutes, 11,000 × g). The aqueous phase containing the nucleotides was collected and stored at -20˚C prior to analysis. Nucleotide analysis was performed using an ESI-TOF mass spectrometer (micrO-TOF II, Bruker) operated in negative-ion mode and connected to an UltiMate 3000 high-performance liquid chromatography (HPLC) system (Dionex). 5 μl of each sample at 10˚C was injected onto the SeQuant ZIC-pHILIC column (Merck, PEEK 150 × 2.1 mm, 5 μm), and the system was run at 30˚C as previously described [43]. The following 40-minute gradient program was used at a flow rate of 0.2 ml/min: 5 minutes of 82% buffer A (CH 3 CN) and 18% buffer B (100 mM (NH 4 ) 2 CO 3 , pH 9.2); 25 minutes of a linear gradient to 42% buffer A; and finally, 10 minutes of 82% buffer A. The DataAnalysis program (Bruker) was used to present the nucleotide masses as extracted-ion chromatograms, and the peak areas were calculated and quantified with Prism 5 (GraphPad). Dilution series of commercially available nucleotides ppGpp (m/z, 601.95), pppGpp (m/z, 681.92) and AMP (m/z, 346.06) were used for calibration to quantify the amounts of nucleotides in the reactions.

Co-IP of Rel Sau with native S. aureus proteins
To generate Rel Sau -specific antibodies, 0.5 mg of purified full-length protein was sent to Davids Biotechnologie GmbH to generate antiserum and affinity-purified IgG. The specificity of the IgG was verified by Western blot analysis (Blot in S1 Fig). For Co-IP, bacteria were grown in 100 ml of CYPG medium to an OD 600 of 1 and centrifuged (5,000 μ g, 5 minutes). The pellet was washed 2 times with PBS and resuspended in 500 μl of cold Co-IP buffer (20 mM HEPES (pH 7.0), 200 mM NaCl, 20 mM MgCl, 20 mM KCl, 0.5 mM DTT, 0.2% (v/v) Tween 20 and cOmplete protease inhibitor cocktail). The resuspended pellet was lysed with 0.5 ml of zirconia-silica beads (0.1 mm diameter) using a high-speed homogenizer (two times, 6,500 rpm, 20 s). Lysed cells were centrifuged for 1 hour at 14,000 x g at 4˚C, and the supernatant was aliquoted (100 μl) and frozen at -80˚C. Co-IP was performed with Dynabeads (Thermo Fisher Scientific) following the manufacturer's instructions with some minor modifications. Briefly, 50 μl of Dynabeads slurry was used for each sample. The storage solution was removed, and the beads were incubated with 30 μg of Anti-Rel Sau IgG resuspended in PBS (pH 7.4) with 0.02% Tween 20 for 30 minutes at room temperature under constant rotation. Coated beads were pelleted using a magnetic rack; the supernatant was removed; and 100 μl of the cell lysates were added and incubated for 30 minutes at room temperature under constant rotation. After incubation, the beads were gently washed 3 times with Co-IP buffer using a magnetic rack. Washing solution was removed, and the beads were resuspended in SDS sample buffer, boiled at 95˚C for 5 minutes, and run approximately 1 cm into an SDS-PAGE gel. The gel slice was subsequently analyzed by mass spectrometry.

Quantitative label-free proteomics
Three biological replicates of (p)ppGpp 0 complemented with WT and mutant Rel Sau were analyzed. Gel slices were digested as described previously [54]. Peptide mixtures were then separated on an EasyLC nano-HPLC (Proxeon Biosystems) coupled to an LTQ Orbitrap Elite mass spectrometer (Thermo Fisher Scientific) as described elsewhere [55] with the following modifications: peptides were eluted with an 87-min segmented gradient of 5-33-90% HPLC solvent B (80% acetonitrile in 0.5% acetic acid). Each sample was run in triplicate. The acquired MS spectra were processed with the MaxQuant software package, version 1.5.2.8 [56] with the integrated Andromeda search engine [57] as described previously [55]. Database searches were performed against a target-decoy S. aureus all-strains database obtained from UniProt, containing 126,225 protein entries and 248 commonly observed contaminants. The label-free algorithm was enabled, as was the "match between runs" option [58]. Label-free quantification (LFQ) protein intensities from the MaxQuant data output were used for relative protein quantification. Downstream bioinformatic analysis (ANOVA and two-sample t-tests) was performed using the Perseus software package, version 1.5.0.15. P < 0.05 was considered to be statistically significant. For the heatmap, among the 4 different proteins, those that showed significant differences according to ANOVA were selected (Data in S1 Dataset). For these selected candidates, the t-test differences, indicating changes in the amount, were calculated between the protein immunoprecipitated with WT or mutant Rel Sau and plotted on the heatmap.

Statistical analysis
The results for the growth and in vitro analyses represent the mean ± SD of at least three biological replicates. Significance was calculated using Prism 5 by one-way ANOVA with Bonferroni correction.